Background
Ischemic stroke is a major cause of mortality and disability. Transient ischemic attack (TIA) is a harbinger of stroke. The etiology of stroke in as many as 40% of patients remains undetermined after extensive evaluation. It was hypothesized that intrapulmonary shunt is a potential facilitator of cerebrovascular accident (CVA) or TIA.
Methods
Patients undergoing clinically indicated transesophageal echocardiography were prospectively enrolled. Comprehensive multiplane transesophageal echocardiographic imaging was performed and saline contrast done to assess for intrapulmonary shunt and patent foramen ovale.
Results
Three hundred twenty-one patients with either nonhemorrhagic CVA ( n = 262) or TIA ( n = 59) made up the stroke group. Three hundred twenty-one age-matched and gender-matched patients made up the control group. Intrapulmonary shunt occurred more frequently in the stroke group (72 of 321) compared with the control group (32 of 321) (22% vs 10%, P < .0001). Intrapulmonary shunt was an independent predictor of CVA and/or TIA (odds ratio, 2.6; P < .0001). In subjects with cryptogenic CVA or TIA ( n = 71), intrapulmonary shunt occurred more frequently (25 of 71) than in the control group (5 of 71) (35% vs 7%, P < .0001). Intrapulmonary shunt was an independent multivariate predictor of CVA or TIA in patients with cryptogenic CVA or TIA (odds ratio, 6.3; P < .005).
Conclusions
These results suggest that intrapulmonary shunt is a potentially unrecognized facilitator of CVA and TIA, especially in patients with cryptogenic CVA and TIA. Future studies assessing the prognostic significance of intrapulmonary shunt on cerebral vascular event recurrence rates in patients after initial CVA or TIA would be of great interest.
Ischemic stroke is a major cause of cardiovascular mortality and long-term disability. Transient ischemic attack (TIA) is a harbinger of stroke. Although many causes of ischemic stroke and TIA are well recognized, as many as 40% of patients after an extensive evaluation are found to have no determined etiology. A cardiac embolic source is believed to potentially explain 25% of these unexplained cases. Echocardiography and transesophageal echocardiography (TEE) in particular have emerged as potentially useful modalities in the evaluation of patients with ischemic cerebrovascular accident (CVA) and TIA. However, several of the findings associated with stroke and TIA as determined by TEE remain controversial, such as patent foramen ovale (PFO), atrial septal aneurysm, and valvular strands. Intracardiac shunt primarily from PFO has been a major focus for a potential explanation of unexplained ischemic stroke. Conversely, limited data exist on the possible role of intrapulmonary shunt as a potential mechanism for ischemic stroke and TIA. Intrapulmonary shunting from pulmonary arteriovenous malformations in patients with hereditary hemorrhagic telangiectasia have been associated with ischemic stroke.
In addition, several case reports of cryptogenic stroke in patients with pulmonary arteriovenous fistula have been reported and suggest that an arteriovenous fistula is a potential facilitator of stroke. To date, no studies have reported in the broader population of patients with strokes or TIAs a potential association with intrapulmonary shunt. The objective of this study was to determine if intrapulmonary shunt is associated with ischemic stroke and TIA.
Methods
Patient Population
Consecutive patients scheduled for clinically indicated TEE from December 16, 2008, to April 14, 2010, were evaluated for study enrollment. Patients who met inclusion and no exclusion criteria were prospectively enrolled. Inclusion criteria were age ≥ 18 years, sinus rhythm, and clinically indicated TEE. Exclusion criteria were hemorrhagic stroke, atrial septal defect, ventricular septal defect, oxygen saturation ≤ 90%, and hemodynamic instability. Patients who underwent TEE more than once over the enrollment time frame were enrolled on only one occasion. Subjects were stratified for recent (≤7 days) nonhemorrhagic CVA and TIA or for other indications. The stroke group included subjects who underwent TEE for recent (≤7 days) nonhemorrhagic CVA or TIA. The control group included patients without prior CVA or TIA. CVA was defined as acute onset of a focal motor, sensory, cranial nerve, speech, or visual deficit as confirmed by examination of a neurologist and lasting ≥24 hours or treated with a thrombolytic. TIA was defined as the abrupt onset of a focal neurologic deficit in the distribution of a major cerebral artery, lasting <24 hours and untreated with thrombolytic therapy. CVA or TIA was subtyped using the Trial of Org 10172 in Acute Stroke Treatment (TOAST) criteria. Cryptogenic CVA or TIA was defined as an event with no definable etiology, including PFO, on the basis of the TOAST criteria. Chronic liver disease was assessed by review of patient records and supporting liver function tests (i.e., serum transaminase, bilirubin, and albumin). The research protocol was approved by our institutional human studies committee. Patients or surrogates provided consent for participation.
TEE
TEE was done with a commercially available echocardiography machine (iE33; Philips Medical Systems, Amsterdam, The Netherlands) using a multiplane probe with harmonic imaging transmitting at 2.8 MHz and receiving at 5.6 MHz. All patients received oxygen supplemental to room air during TEE. The amount of supplemental oxygen was left to the discretion of the echocardiographer performing the exam. After insertion of the probe, a comprehensive two-dimensional and color and spectral Doppler transesophageal echocardiographic examination was performed. Ten milliliters of saline with 0.1 to 0.2 mL of air vigorously agitated between two syringes with a three-way stopcock was injected in an arm vein as a contrast medium to assess for intrapulmonary shunt and patency of the foramen ovale membrane. Saline contrast studies were done without provocative maneuvers, such as cough or Valsalva maneuver, to simplify differentiation of late bubbles entering the left atrium on the basis of intrapulmonary shunt from those due to provoked opening of an intermittent PFO.
Lack of a provocative maneuver during saline contrast TEE allowed a definitive differentiation of the path by which bubbles entered the left atrium (i.e., either by traversing the atrial septum or entering through the pulmonic veins). Appearance of bubbles in the left atrium during a provocative maneuver is not specific for a PFO.
The left superior and inferior pulmonic veins and alternatively the right superior and inferior pulmonic veins were imaged simultaneously from a longitudinal plane. A horizontal plane was used if necessary for adequate visualization of pulmonic veins. Four to five saline injections were typically needed to assess for intrapulmonary shunting via the pulmonic veins and patency of the foramen ovale membrane. Twenty-beat cardiac cycle digital acquisitions were acquired with each contrast injection.
Intrapulmonary shunt was defined as direct visualization of bubbles entering the left atrium from one or more pulmonic veins. PFO was defined as bubbles visualized to traverse the atrial septum or alternatively to enter the left atrium within three cardiac cycles of opacification of the right atrium. A mild, intermediate, or large intrapulmonary shunt was defined as one to four, five to 20, or >20 bubbles on any single frame within the pulmonic vein(s) and entering the left atrium from the pulmonic vein(s) after a single contrast injection. In the case of shunt from multiple pulmonic veins, the degree of shunt was based on the pulmonic vein yielding the highest severity. A mild, intermediate, or large PFO was defined as one to four, five to 20, or >20 bubbles on any single frame entering the left atrium via the atrial septum with any single contrast injection. The criteria for the severity of an intrapulmonary shunt were extrapolated from similar criteria used to grade PFO. Care was taken to discern bubbles shunting from the right upper pulmonic vein from those traversing a PFO or bubbles retrogradely flowing into the pulmonic veins after entering the left atrium through a PFO. Left ventricular ejection fraction was measured using TEE by the modified Simpson’s method or qualitatively if the left ventricle was deemed to be foreshortened. Right ventricular (RV) size and function were assessed from a four-chamber view. RV area > 24 cm 2 was defined as RV enlargement, and RV fractional area change <32% was defined as RV dysfunction. Echocardiographic studies were analyzed in a blinded fashion.
Statistical Analysis
All values represent mean ± SD or proportions of groups. Unpaired Student’s t tests were used to compare continuous variables. Fisher’s exact tests were used to assess for differences in proportions. Mann-Whitney U tests were used to assess for differences in proportions of the severity of intrapulmonary shunt, number of pulmonic veins demonstrating intrapulmonary shunt, and severity of PFO. Multivariate logistic regression analysis was used to determine predictors of CVA and TIA and of intrapulmonary shunt. Odds ratios (ORs) and 95% confidence intervals (CIs) were calculated. P values <.05 were considered statistically significant. All statistical analysis was performed using SPSS version 17.0 (SPSS, Inc., Chicago, Illinois).
Results
Clinical and Transesophageal Echocardiographic Findings among Groups
Nine hundred fifty-four transesophageal echocardiographic studies were done in 891 patients over the enrollment period. The 63 repeat studies of patients who had already been enrolled were not included in the analysis. Twelve patients met exclusion criteria, three patients refused enrollment, and in two patients the probe could not be passed. Thus, 874 patients meeting inclusion and no exclusion criteria made up the study cohort. Of these patients, 523 subjects (280 men, 243 women) had either nonhemorrhagic CVAs ( n = 424) or TIAs ( n = 99). The remaining 351 patients (230 men, 121 women) had no prior CVA or TIA. From this cohort, patients with CVAs or TIAs were then matched one to one by gender and age within 3 years with subjects with no prior CVA or TIA. Three hundred twenty-one patients (204 men, 117 women) with recent CVAs ( n = 262) or TIAs ( n = 59) made up the stroke group. Three hundred twenty-one patients (204 men, 117 women) with no prior CVA or TIA made up the matched control group. The mean ages of the stroke group and the control group did not differ ( Table 1 ). The clinical indications for TEE in the control group were endocarditis ( n = 98), valvular heart disease ( n = 107), nonatrial arrhythmias ( n = 24), heart failure ( n = 57), hypotension ( n = 2), suspected aortic dissection ( n = 8), suspected intracardiac mass ( n = 9), pulmonary hypertension ( n = 2), pseudoaneurysm ( n = 1), pericardial mass or disease ( n = 2), congenital heart disease ( n = 3), suspected pulmonary embolism ( n = 1), myocardial infarction ( n = 1), contusion ( n = 1), cardiomegaly ( n = 1), coronary anomaly ( n = 1), and miscellaneous ( n = 3). Race and diabetes did not differ between the groups ( Table 1 ). Hypertension, hyperlipidemia, and current or prior tobacco use were more frequent in the stroke group ( Table 1 ). In the stroke group, identifiable causes of CVA or TIA were found in 250 subjects ( Table 1 ). High-risk aortic plaque (≥5 mm) or mobile thrombus in the ascending aorta and/or aortic arch were uncommon but were found more frequently in the stroke group ( n = 18) than in the control group ( n = 7) ( P < .05; Table 1 ). Cryptogenic CVA or TIA occurred in 71 patients by the TOAST criteria ( Table 1 ). The mean left ventricular ejection fraction was higher in the stroke group (60 ± 11%) than in the control group (53 ± 16%) ( P < .0001). RV enlargement occurred less frequently in the stroke group (22 of 321) compared with the control group (49 of 321) (7% vs 15%, P < .0001). RV dysfunction occurred less frequently in the stroke group (13 of 321) compared with the control group (63 of 321) (4% vs 20%, P < .005).
| Variable | Stroke group ( n = 321) | Control group ( n = 321) | P |
|---|---|---|---|
| Age (y), mean ± SD | 55.86 ± 13.03 | 55.81 ± 13.02 | NS |
| Men | 204 (64%) | 204 (64%) | NS |
| Race | |||
| White | 238 (74.2%) | 221 (68.8%) | |
| Black | 68 (21.2%) | 86 (26.8%) | |
| Hispanic | 2 (0.6%) | 0 (0%) | NS |
| Asian | 9 (3%) | 9 (2.8%) | |
| Other | 4 (1%) | 5 (1.6%) | |
| Tobacco use | 198 (62%) | 156 (49%) | <.005 |
| Diabetes | 99 (31%) | 104 (32%) | NS |
| Hypertension | 256 (80%) | 203 (63%) | <.0001 |
| Hyperlipidemia | 171 (53%) | 91 (28%) | <.0001 |
| Etiology of CVA or TIA based on TOAST criteria | NA | NA | |
| 1. Large vessel | 97 (30%) | ||
| 2. Small vessel | 35 (11%) | ||
| 3. Cardioembolic | 111 (35%) | ||
| 4. Others | 7 (2%) | ||
| Moyamoya | 2 | ||
| Lupus vasculitis | 3 | ||
| Thrombophilia | 1 | ||
| Cocaine induced vasospasm | 1 | ||
| 5. Cryptogenic | 71 (22%) | ||
| Antihypertensive medications at baseline | 188 (59%) | 183 (57%) | NS |
| Lipid-lowering medications at baseline | 113 (35%) | 94 (29%) | NS |
| Anticoagulant medications at baseline | 22 (7%) | 35 (11%) | NS |
| Antiplatelet medications at baseline | 132 (41%) | 103 (32%) | <.05 |
| Antidiabetic medications at baseline | 75 (23%) | 79 (25%) | NS |
| High-risk plaques (severe plaques and/or mobile thrombi in ascending aorta or arch) | 18 (5.6%) | 7 (2.2%) | <.05 |
| Severe plaque and/or mobile thrombus in ascending aorta | 4 (1.3%) | 1 (0.3%) | NS |
| Severe plaque and/or mobile thrombus in aortic arch | 17 (5.3%) | 7 (2.2%) | NS |
Intrapulmonary Shunt
Intrapulmonary shunt occurred more frequently in the stroke group (72 of 321) compared with the control group (32 of 321) (22% vs 10%, P < .0001). In patients with strokes ( n = 72) who had intrapulmonary shunts, the severity of the shunts was at least intermediate in the majority of subjects ( Table 2 , Figures 1–3 , Video 1 ; available at www.onlinejase.com ). Similar findings were seen in control patients who had intrapulmonary shunts ( Table 2 ). The distribution of severity of intrapulmonary shunt differed ( P < .0001) between these groups ( Table 2 ). In patients with strokes ( n = 72) who had intrapulmonary shunts, a majority had shunting from multiple pulmonic veins ( Table 2 ). Similar findings were seen in the control group ( Table 2 ). The distribution of the number of pulmonic veins accounting for shunting differed between these groups ( P < .0001). In the stroke group ( n = 321), oxygen supplementation of ≥40% ( n = 30) compared with <40% ( n = 291) showed no significant difference in frequency of intrapulmonary shunt (22% vs 23%, P = NS). In the control group ( n = 321), oxygen supplementation of ≥40% ( n = 131) compared with <40% ( n = 190) did not show any difference in the frequency of intrapulmonary shunt (8% vs 12%, P = NS). Oxygen supplementation of ≥40% was less common in the stroke group (30 of 321) than in the control group (131 of 321) (9% vs 41%, P < .0001). Supplemental 100% oxygen was administered to 21 of 321 patients in the stroke group, compared with 39 of 321 patients in the control group (6.5% vs 12.2%, P < .05). The frequency of intrapulmonary shunt in patients administered 100% oxygen did not differ between subjects in the stroke (four of 21) and control (five of 39) groups (19% vs 13%, P = NS). Intrapulmonary shunt occurred in 35% of subjects (25 of 71) with cryptogenic CVAs or TIAs and in 19% of subjects (47 of 250) with identified potential causes of CVAs or TIAs ( P < .005). Chronic liver disease occurred in only 4.8% of subjects (5 of 104) with intrapulmonary shunts.
| Variable | Stroke subgroup with intrapulmonary shunt ( n = 72) | Control subgroup with intrapulmonary shunt ( n = 32) | P |
|---|---|---|---|
| Severity of intrapulmonary shunt | |||
| Mild | 22 (31%) | 8 (25%) | |
| Intermediate | 49 (68%) | 22 (69%) | |
| Large | 1 (1%) | 2 (6%) | <.0001 |
| Number of pulmonic veins with shunts | |||
| 1 | 21 (29%) | 6 (19%) | |
| 2 | 21 (29%) | 15 (47%) | |
| 3 | 3 (4%) | 2 (6%) | |
| 4 | 27 (38%) | 9 (28%) | <.0001 |
PFO
The frequency of PFO in the stroke group (73 of 321) was slightly but not significantly higher than that in the control group (54 of 321) (23% vs 17%, P = .07). In subjects ( n = 73) with PFOs in the stroke group, 32 (44%), 26 (36%), and 15 (20%) had mild, intermediate, and large shunts ( Video 2 ; available at www.onlinejase.com ), respectively. In subjects ( n = 54) with PFOs in the control group, 26 (48%), 23 (43%), and five (9%) had mild, intermediate, and large shunts, respectively. The presence of a PFO and/or an intrapulmonary shunt was more common in the stroke group (136 of 321) than in the control group (84 of 321) (42% vs 26%, P < .0001).
Predictors of Stroke or TIA and Intrapulmonary Shunt
Multivariate logistic regression analysis was performed to determine if intrapulmonary shunt was a predictor of stroke and/or TIA. Clinical variables were age, gender, diabetes, race, hypertension, tobacco abuse, and hyperlipidemia. Echocardiographic variables were left ventricular ejection fraction, RV dysfunction, intrapulmonary shunt, and PFO. Hypertension (OR, 1.9; 95% CI, 1.3–2.8; P < .005), hyperlipidemia (OR, 2.1; 95% CI, 1.5–3; P < .0001), RV dysfunction (OR, 0.17; 95% CI, 0.09–0.33; P < .0001), presence of intrapulmonary shunt (OR, 2.6; 95% CI, 1.6–4.2; P < .0001), and PFO (OR, 1.7; 95% CI, 1.1–2.6; P < .05) were independent predictors of CVA and/or TIA ( R = 0.46, P < .0001). Clinical variables (i.e., age, gender, race, chronic liver disease, tobacco abuse, and oxygen supplementation ≥40%) and echocardiographic parameters (i.e., RV size and RV dysfunction) did not predict intrapulmonary shunt ( P = NS).
Cryptogenic Stroke or TIA
A subgroup of patients ( n = 71) with cryptogenic strokes ( n = 50) or cryptogenic TIAs ( n = 21) and their respective age-matched and gender-matched controls ( n = 71) were analyzed as a separate cohort. Race, hypertension, current or prior tobacco abuse, and diabetes did not differ among the groups ( Table 3 ). The mean left ventricular ejection fraction was higher in the cryptogenic stroke group (62.7 ± 6.2%) than in the cryptogenic control group (53.7 ± 17.4%) ( P < .0001). RV enlargement occurred less frequently in the cryptogenic stroke group (zero of 71) compared with the cryptogenic control group (11 of 71) (0% vs 16%, P < .005). RV dysfunction occurred less frequently in the cryptogenic stroke group (zero of 71) compared with the cryptogenic control group (13 of 71) (0% vs 18%, P < .0001).
| Variable | Cryptogenic stroke group ( n = 71) | Control group ( n = 71) | P |
|---|---|---|---|
| Age (y), mean ± SD | 52.28 ± 11.42 | 52.27 ± 11.43 | NS |
| Men | 44 (62%) | 44 (62%) | NS |
| Race | |||
| White | 56 (78.9%) | 49 (69%) | |
| Black | 13 (18.3%) | 22 (31%) | |
| Hispanic | 0 (0%) | 0 (0%) | NS |
| Asian | 1 (1.4%) | 0 (0%) | |
| Other | 1 (1.4%) | 0 (0%) | |
| Tobacco use | 44 (62%) | 33 (47%) | NS |
| Diabetes | 20 (28%) | 21 (30%) | NS |
| Hypertension | 50 (70%) | 41 (58%) | NS |
| Hyperlipidemia | 33 (47%) | 16 (23%) | <.005 |
Stay updated, free articles. Join our Telegram channel
Full access? Get Clinical Tree
